Biochemistry 1991, 30, 3371-3386 Omichinski, J. G., Clore, G. M., Appella, E., Sakaguchi, K., & Gronenborn, A. M. (1 990) Biochemistry 29,9324-9334. Parraga, G . , Horvath, S., Eisen, A., Taylor, W. E., Hood, L., Young, E. T., & Klevit, R. E. (1988) Science 241, 1489-1492. Parraga, G., Horvath, S., Young, E. T., & Klevit, R. E. (1990) Proc. Natl. Acad. Sci. U.S.A. 87, 137-141. Rance, M., Sorensen, 0. W., Bodenhausen, G., Wagner, G.,
3371
Ernst, R. R., & Wuthrich, K. (1983) Biochem. Biophys. Res. Commun. 117, 479-485. Weiss, M. A,, & Keutmann, H. (1990) Biochemistry 29, 9808-98 13. Weiss, M. A,, Mason, K. A., Dahl, C. E., & Keutmann, H. ( 1990) Biochemistry 29, 5660-5664. Wuthrich, K. (1986) N M R of Proteins and Nucleic Acids, Wiley, New York.
Alternating Zinc Fingers in the Human Male Associated Protein ZFY: 2D NMR Structure of an Even Finger and Implications for "Jumping-Linker" DNA Recognition? Michel Kochoyan,fs Timothy F. Havel,* Dzung T. Nguyen," Charles E. Dahl,* Henry T. Keutmann," and Michael A. Weiss*,*,ll Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts 021 15, and Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts 021 14 Received October 1 7, 1990; Revised Manuscript Received December 26, I990
a sex-related Zn-finger protein encoded by the human Y chromosome, is distinguished from the general class of Zn-finger proteins by the presence of a two-finger repeat. Whereas odd-numbered domains and linkers fit a general consensus, even-numbered domains and linkers exhibit systematic differences. Because this alternation may have fundamental implications for the mechanism of protein-DNA recognition, we have undertaken biochemical and structural studies of fragments of ZFY. W e describe here the solution structure of a representative nonconsensus (even-numbered) Zn finger based on 2D N M R studies of a 30-residue peptide. Structural modeling by distance geometry and simulated annealing (DG/SA) demonstrates that this peptide folds as a miniglobular domain containing a C-terminal @-hairpinand N-terminal a-helix (PPa motif). These features are similar to (but not identical with) those previously described in consensus-type Zn fingers (derived from A D R l and Xfin); the similarities suggest that even and odd Z F Y domains bind D N A by a common mechanism. A model of the protein-DNA complex (designated the "jumping-linker" model) is presented and discussed in terms of the Z F Y two-finger repeat. In this model every other linker is proposed to cross the minor groove by means of a putative finger/linker submotif HX,HX,-hydrophobic residue-X3. Analogous use of a hydrophobic residue in a linker that spans the minor groove has recently been described in crystallographic and 3D N M R studies of homeodomain-DNA complexes. T h e proposed model of Z F Y is supported in part by the hydroxyl radical footprint of the TFIIIA-DNA complex [Churchill, M. E. A,, Tullius, T. D., & Klug, A. (1990) Proc. Nutl. Acad. Sci. U.S.A. 87, 5.528-55321. ABSTRACT: ZFY,
T b e Zn-finger motif defines a highly conserved class of eukaryotic nucleic acid binding proteins (Klug & Rhodes, 1987; Evans & Hollenberg, 1988). In this paper we describe 2D NMR' studies and structural modeling of an isolated Zn finger from the human male associated protein ZFY. This gene, originally identified from studies of sex reversal in man (de 'Supported by the Lucille Markey Charitable Trust and by grants From the National Institutes of Health (HD 26465), the American Cancer Society, and the Whitaker Foundation to M.A.W. M.A.W. is supportcd i n part by the Pfizer Scholars Program for New Faculty and a Junior Faculty Research Award from the American Cancer Society. T.H. is supported in part by NIH Grant GM 38221. M.K. is supported in part by a NATO postdoctoral fellowship. * Addrcss correspondence to this author at the Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School. 8 Harvard Medical School. Permanent address: CNRS URA 1254, BlOP Polytechnique, 91 128 Palaiseau. France. I' Massachusetts General Hospital.
0006-2960/9 1/0430-337 1$02.50/0
la Chapelle, 1972; Page et al., 1987), encodes a putative transcription factor proposed to participate in spermatogenesis (Palmer et al., 1989; Koopman et al., 1989). Genetic studies of intersex abnormalities of the newborn suggest that ZFY also plays an accessory role in the pathway of male sexual development in embryogenesis (Page et al., 1990). Two-dimensional N MR studies have previously been conducted of single-finger peptides from Saccharomyces protein ADRl and Xenopus protein Xfin (Parraga et al., 1988; Lee ct al., 1989a,b). These fingers fold as compact globular minidomains in which the divalent metal is encaged. The Nterminal portion of the finger, containing the conserved cysteines, forms a @-sheetand @-turn(@-hairpin);the C-terminal
' Abbrcviations: CD, circular dichroism; DQF-COSY, double-quantum-filtered correlated spectroscopy; DG, distance geometry; NMR, nuclcar magnetic resonance; NOESY, nuclear Overhauser effect spectroscopy; RMS, root mean square; SA, simulated annealing; TOCSY, total correlation spectroscopy. 0 1991 American Chemical Society
3372 Biochemistry, Vol. 30, No. 14, 1991 portion, containing the conserved histidines, forms an a-helix (PO. motif). These features are in accord with earlier structural models (Berg, 1988; Gibson et al., 1988). The putative DNA-binding domain of ZFY (Page et al., 1987) is distinguished from the general class of Zn-finger proteins by the presence of a two-finger repeat (Figure I ) . In this repeat the odd-numbered domains (fingers 1, 3, 5, 7 , 9 , 11, and 13) fit the general Zn-finger consensus sequence (Gibson et al., 1988), whereas the even-numbered domains (fingers 2 , 4 , 6, 8, IO, and 12) contain systematic differences. Circular dichroism studies suggest that peptides derived from odd- or even-numbered domains exhibit similar but not identical patterns of metal-dependent folding (Weiss et al., 1990). In this paper we will focus on the solution structure of a representive even-numbered domain. Our results are presented in two parts. Part 1 describes the design of a 30-residue peptide (designated ZFY-6T)2 as a model system and the sequential assignment of its 2D N M R spectrum. The observed pattern of nuclear Overhauser enhancements (NOEs) and J-coupling constants demonstrates that this variant finger retains the 0-sheetla-helix motif as previously described (Parraga et al., 1988; Lee at al., 1989a,b; Klevit et al., 1990). Long-range NOEs involving conserved framework residues define a hydrophobic core, and observation of slowly exchanging amide resonances indicates that this core is exceptionally stable. The three-dimensional structure of ZFY-6T is obtained in part 11 by analysis of N O E and J coupling restraints using distance geometry and simulated annealing (DG/SA). The overall pattern of folding is similar to consensus-type Zn fingers; well-defined interactions are observed involving residues conserved among Zn fingers or particular to the subfamily of ZFY-related sequences. Our results suggest that odd- and even-numbered fingers share a common mechanism of DNA binding. A model of the ZFY protein-DNA complex (designated the "jumping-linker" model) is presented that retains the two-finger repeat as an essential feature. Aspects of this model are supported by a reinterpretation of the high-resolution hydroxyl radical footprinting of the TFIITA-DNA complex (Churchill et al., 1990) and discussed in reference to analogous features of the homeodomain-DNA complex (Otting et al., 1990; Kissinger et al., 1990). MATERIALS A N D METHODS Peptide Synthesis and Characterization. ZFY-6P (sequence KPYQCQYCEYRSADSSNLKTHIKTKHSKEK) and ZFY-6T (sequence KTYQCQYCEYRSADSSNLKTHIKTKHSKEK) were synthesized by the solid-phase procedure (Barany & Merrifield, 1979; Stewart & Young, 1984) and purified following reduction (below) by reversed-phase HPLC as previously described (Weiss et al., 1990). Quantitative ninhydrin tests were used to monitor coupling efficiency, which averaged 99.1-99.376. The following protecting groups were used on the t-BOC amino acids: chlorobenzoxy (Lys), carbobenzoxy (His), bromobenzoxy (Tyr), methylbenzyl (Cys), benzyl (Ser, Thr, Glu, Asp), and tosyl (Arg). Purity was evaluated by analytical HPLC, composition, and sequence
Nomenclature: In this paper ZFY-6P denotes the 30-mer peptide containing proline at position 2, ZFY-6T denotes the corresponding pcptidc with threonine at position 2, and ZFY-6 denotes in generic terms the even-specific motif of domain 6. Odd-numbered linkers are defined as those that follow odd-numbered ZFY domains; even-numbered linkers are defined as those that follow even-numbered ZFY domains. Amino acids are designated by standard single-letter code.
Kochoyan et al. analysis of preview (Tregear et al., 1977). Peptides were also prepared by using F-moc methodology as previously described (Weiss et al., 1990). Reduction of Cystine to Cysteine. This was accomplished by reaction with 0.5 M dithiothreitol in 100 mM Tris-HCI (pH 7.7 at 20 "C) at 60 "C for 1-3 h (Frankel et al., 1987). The reduction status of cysteines was confirmed by reaction with iodoacetate followed by sequencing. The reduced peptide exhibited a different elution position in reversed-phase HPLC than an oxidized form containing an intramolecular disulfide bond, thus enabling the reduced and oxidized forms to be characterized individually (Weiss & Keutmann, 1990). Aggregation State. This was determined by gel filtration chromatography (Sephadex (3-50 Fine). The elution position cxpected for a monomeric peptide was calibrated in reference to a fragment of parathyroid hormone (residues 1-34). The oligomeric states of the ZFY-6P/Zn2+ and ZFY-6T/Zn2+ complexes were found to be monomeric a t the concentration and conditions of study (see part I under Results). N M R Sample Preparation. A total of 12 mg of the reduced pcptide was dissolved in 0.7 mL of deoxygenated N M R buffer (see below). The concentration of peptide was ca. 5 mM. Following N M R data collection, the sample was lyophilized and stored in vacuo; under such conditions the ZFY-6T/Zn2+ complex did not undergo significant oxidation ( 5 . 5 a single spin system is observed in the fingerprint region for each residue (Figure 5; see below). This is consistent with the existence of a single conformation of the metal-folded peptide (an apparent minor conformer of the Y10 side chain is discussed below). The structure remains folded up to 85 "C (panels B-D). Similar metal-dependent 'H NMR changes are observed upon addition of Cd2+ (data not shown). pH-Dependent Unfolding. Coordinate loss of metal-binding and secondary structure has previously been described in the ZFY-6P/Co2+ complex between pH 5 and 6 (Weiss et al., 1990; Weiss & Keutmann, 1990); similar pH behavior has been observed in other systems (Parraga et al., 1990). In CD studies the ZFY-6P/Zn2+ complex exhibits similar loss of metal-dependent structure between pH 4 and 5; the shift in pH midpoint reflects the greater stability of the Zn2+complex relative to Co2+. pH-dependent folding and unfolding may also be monitored by ' H NMR. Spectra obtained at the midpoint of the pH-unfolding curve demonstrate slow exchange between folded and unfolded species, reflecting a two-state process (Figure 3). In the NOESY spectrum exchange cross peaks are observed between liganded (H26 and H21 at 7.44 and 7.84 ppm, respectively) and unliganded (8.6 ppm) histidine C,H resonances (labeled t l and c2 in Figure 3).
-I-
;o
] 0
0
-1 I FIGURE 5: daN(i,i+l)connectivities of ZFY-6T in the NOESY
spectrum recorded in 90% H20(conditions: 250-ms mixing time, pH 6.2, and 25 "C). Cross peaks are labeled by residue number. Helix-related daN(i,i+3)connectivities in the C-terminal domain are indicated by arrows (see Figure 7).
6.5 -
.
..
e
7.0 7.5 -
8.0 8.5 9.0IO
1
I
I
I
I
I
9.5
9.0
8.5
80
7.5
I
I
7.0 6.5 6
FIGURE 6: dNN(i,i+l)connectivities of ZFY-6T in the NOESY spectrum recorded in 90% H 2 0 (conditions: 250-ms mixing time, pH 6.2, and 25 "C). Cross peaks are labeled by residue number. The arrow indicates an interstrand NOE between residues 3 and 12 in the P-sheet.
As expected, NOES characteristic of the folded state-as between, for example, the CBHprotons of the liganded histidines (cross peak e) or between the ring protons of Y10 (unresolved at 6.42 ppm; see below) and the C,H proton of H21 (cross peak h in Figure 3)-are retained under these conditions. Two-state behavior has also been described in the pH-dependent unfolding of A D R l a (Parraga et a]., 1990). N M R Spin-System Identification. ZFY-6P and ZFY-6T each contain 30 residues. Their constituent spin systems may be classified as (i) methyl containing (A, I, L, and T), (ii) long methylene chains (K, R, Q, and E), (iii) aromatic residues (Y, H), and (iv) remaining AMX spin systems (C, S, N, and D). In this section these spin systems in ZFY-6T are described in turn; the spectrum of ZFY-6P is similar and will be described below. ( i ) Methyl-Containing Spin Systems. ZFY-6T contains three threonines (positions 2, 20, and 24) and unique alanine (position 13), leucine (1 8), and isoleucine (22) residues. Their characteristic spin systems are readily identified in DQFCOSY and TOCSY spectra. (ii) Long Side Chains. Eleven out of the 30 residues of ZFY-6T have long side chains. The six lysines (positions 1, 19, 23, 25, 28, and 30) and unique arginine (position 11) are readily identified in the TOCSY spectrum (Figure 4); in each
Kochoyan et al.
3376 Biochemistry, Vol. 30, No. 14, 1991 1 K
2 T
3 Y
4 Q
5 C
6 Q
7 Y
8 C
9 E
10 Y
I 1 12 R S
13 R
14 D
15 16 17 18 19 20 2 1 2 2 23 2 4 25 26 27 28 29 3 0 S S N L K T H I K T K H S K E K
-
daN( 1 I 1 t 4 )
4 -
d (i,it3) PN 0
.
0
0
8
0 0 0 0 0 0 0 0
0 0
3J HNa
0 5 , s Hz 8.5 Hz < 0 FIGURE 7: Schematic representation of the sequential connectivities and backbone J-coupling constants of ZFY-6T. Symbols and format are as described (Wuthrich, 1986).
q
iI i
NH
H
/
R'
82
H=
I 2.2
2.0
1.8
1.6
I4
1.2
1.0
I
1
1
I
1
I
9 5
90
85
80
75
70
case connectivities are observed from H, to each successive proton, including 6 protons of arginine (cross peak h in panel B of Figure 4) and 6 protons of lysine (cross peaks a-e). Four additional spin systems are observed due to glutamine or glutamic acid, which cannot be further distinguished a t this stage of the analysis. (iii) Aromatic Residues. The aromatic region of ZFY-6T is well resolved. The ortho and meta resonances of two of the three tyrosines are classified on the basis of their connectivities in the DQF-COSY and NOESY spectra. The H, protons of thc two histidine in the folded peptide are assigned from their cxchange cross peak with the same protons of the unfolded peptide a t pH 4.5 (see above). The H6 protons are then assigned from their strong TOCSY and weak intraresidue N O E connectivity to the H,. An additional composite resonance (6.4 ppm) is observed i n D 2 0 , which upon integration contains approximately four proton resonances; this is assigned to the
1
I
6 5
PPm
ppm FiGuRE 8: Stereospecificassignment of L18 obtained by an extension of the method of Wagner et al. (1987) for P-protons. Following stereospecificassignment of the P protons (summarized in a schematic form in the lower left panel), additional intraresidue NOES [Le.,strong NOE cross peaks from H, to H, and C,,H, (labeled Me6'), from C6*H3 (labeled Me6*in the figure) to both P protons, and from H,, to both n x t h y l resonances] make possible the stereospecific assignment of the LI 8 methyl groups (summarized in a schematic form in the upper lcft pancl; sce also text). Experimental conditions: 100-ms mixing time, pH 6.2, and 25 OC.
Id
Slowly exchanging amide resonances of ZFY-6T at pD 6.0 (direct meter reading) demonstrating the stability of the Zn finger as a globular minidomain. Spectrum A was recorded 15 min after dissolution of a protonated sample in D20. Assignments are indicated (see Table I). Progressive exchange is observed over 24 h under these conditions. Spectrum B is a reference spectrum of the same sample FIGURE 9:
in 90'X H20/10%)D 2 0 .
unresolved ortho and meta protons of the third tyrosine. In each case the histidine H6and tyrosine ortho resonances exhibit strong NOES to the protons of a single AMX spin system, enabling assignment of the residue side chains to be completed. ( i r ) Other A M X Spin Systems. Eight remaining AMX spin systems (corresponding to four Ser, two Cys, one Asp, and one Asn; see Table I) are identified but not further distinguished a t this stage. Sequential Assignment. The N H-C,H fingerprint region is well resolved in the NOESY spectrum of the Zn2+ complex (Figure 5). C,H, to NH,,, connectivities are observed from residues 2 to 14 and from residues 16 to 30. In addition, amide-amide connectivities are observed from residues 6 to 1 1 and 16 to 27, as outlined in Figure 6. This network of connectivities provides the sequential assignment of residues 3 to 14 and 16 to 30. At pH 6-6.5 the N H resonances of T2 and S 15 are exchanging too rapidly with water to give rise to observable resonances: these resonances are assigned on the basis of a DQF-COSY spectrum recorded a t p H 5.5. The N-terminal N H 2 resonance of K1 is not observed under these
Biochemistry, Vol. 30, No. 14, 1991 3317
"Jumping-Linker" DNA Recognition
Q
YIO-
H26L6)Y7/3(melo)
0
c
-
0
-
-
;=.=
I
Y 7 ( o r t h o )-
=-66
a43 0
C
-6 8
0
Y 3 ( o r t h o )-
H21 ( 8 ) -
-64
OQ 0
0
Q
=
0
-7 0
Q
-7 2
H2l(6)-
0
c
Q
Q
-74
=
-76
-
H26(6)I
I
I
1
I
I
= I
I
-78 I
FIGURE 10: NOESY cross peaks between the aromatic protons of the histidines and tyrosines (vertical axis; a,)and methyl protons (horizontal axis; w 2 ) demonstrating tertiary interactions in the hydrophobic core of the domain. Assignments are indicated (see Table I). The mixing time was 200 ms. Stereospecific NOEs indicate the absence of significant spin diffusion under these conditions.
conditions. A summary of sequential connectivities is shown in schematic form in Figure 7; ' H N M R assignments of ZFY-6T are given in Table I. The assignment of two AMX spin systems as ligands (C5 and C8) is verified by observation of heteronuclear '13Cd coupling (data not shown). StereospeciJic Assignments. Stereospecific assignments have been obtained for the @ protons of 10 residues on the basis of (i) the value of the C,H-C,H coupling constants, (ii) the relative intensities of the intraresidue C,H-C,H NOEs, and (iii) the relative intensities of the NH-C,H NOEs. Stereospecific assignments have been used for those residues for which all criteria are compatible with one of the three preferred side-chain rotamers (Wagner et al., 1987). No assignments are made for the residues for which one of the coupling constants is in the range of 6-9 Hz or for which intraresidue NOE intensities are incompatible with the apparent coupling constants. Once the stereospecific assignments of the @ protons have been obtained, it is possible, in principle, to extend the previous procedure to the chemical groups attached to the C, carbon. This has been done in the case of Leu18 (Figure 8). The stereospecific assignment of the two 6-CH3 resonances is based on their respective intraresidue N O E and coupling patterns: (i) strong NOES from 61-CH3 to C,H and CB2Hprotons and absent NOES from 61-CH3 to CB3H,(ii) strong NOEs from 62-CH3 to C,,H and C,,H protons and absent NOEs from 62-CH3 to C,H, and (iii) observation of a large coupling constant ( > l o Hz) between C,H and C,,H. (Due to resonance overlap the coupling constant between C,H and C,,H could not be obtained.) Rotamer relationships underlying this approach are depicted in schematic form in Figure 8. Secondary Structure. The analysis of sequential and intermediate N O E patterns and of the NH-C,H J-coupling constants (Figure 7) enables two regions in the structure of the finger to be distinguished: an N-terminal &sheet and @-turn (P-hairpin) and a C-terminal a-helix (@Pamotif). These elements of secondary structure are discussed in turn. (i) N-Terminal @-Sheet. The N-terminal region, from residue 2 to residue 13, is characterized by two subdomains with strong daNN O E connectivities (2 to 6 and 8 to 12) and by strong 'JHNa coupling constants in residues 3,4, 7, and 11 (summarized in Figure 7). These features suggest the presence
of two strands of extended chains. Interstrand NOES observed between residues 2, 3, and 4 and residues 13, 12, and 11 (respectively) in turn indicate formation of a &hairpin; the position of the turn is localized by dNNN O E connectivities between residues 6 and 8. A representative interstrand NOE (between the H N resonances of Y3 and S12) is indicated by an arrow in Figure 6. (ii) C-Terminal a-Helix. The C-terminal domain between residues 15 and 26 exhibits an N O E pattern [dm, d,~(i$+3), and daN(i,i+4)] (Figure 7) characteristic of an a-helix. Additional evidence for a-helix formation is provided by small 'JHN, coupling constants in residues 16 to 23 (
